The Dunning-Kruger effect may just be a data artefact (2020)
https://www.mcgill.ca/oss/article/critical-thinking/dunning-kruger-effect-probably-not-realOpen linkView original on piefed.zipSyndicated from the fediverse. Read and engage on the original instance.
View original on piefed.ziphttps://www.mcgill.ca/oss/article/critical-thinking/dunning-kruger-effect-probably-not-realOpen linkView original on piefed.zip
8 replies
This thesis comes from a study that correctly attributes Dunning-Kruger‘s metacognitive gap to “statistical noise,” which is why the author claims that the effect is an illusion.
He’s not even wrong, and far from invalidating the very intuitive, very obviously correct DK effect, the study just explains it.
Greater familiarity with your own abilities is part of skill acquisition, so when you guess your “score,” you do so with more precision. More importantly, however, because you’re in the upper bell curve, random guesses have much more room when you underestimate your performance than when you overestimate it, even if you’re doing so with equal frequency.
Someone who sucks, on the other hand, has way more room to overestimate their performance than to underestimate it.
For instance, if I claimed that my LSAT score would be perfect, I’d still be accurate to within 5 points, so my overconfidence hardly seems like hubris.
Compare that to a business major I knew who guessed he would score in the 75th percentile only to end up in the 5th percentile. The gap between his mediocre expectations and the brutal reality of being illiterate was substantial. And yet, it almost seems unreasonable to expect him to guess accurately! What is he supposed to say, “I’m gonna get a zero”?
The point is, the DK effect is a statistical artifact, and that changes absolutely nothing.
The statistical artifact is the interesting part so many people reference, so I don't see the problem. It's just often used to reference only the subset of idiots that don't know much and think they do. That's still a big section of people exhibiting peculiar behavior even if the conclusion is, "on average, people over and under estimate at similar rates". It was never the about the rates.
I don't see how this does not invalidate the original finding. The original was a statistical study based on an incorrect methodology which produced incorrect results and interpretation.
How can you claim the effect does exist if it has not been measured?
Indeed I do observe at times people who do not know much of a subject overestimate their abilities and sometimes I see people who are experts underestimating their abilities.
This does not however rule out that in many other occasions people who are inexperienced underestimate their abilities and people who are experts overestimate their abilities.
The first rule of Dunning-Kruger club is that you don't know you are in Dunning-Kruger club.
The dunning-kruger effect also is effected by the dunning-kruger effect. Which is both unfortunate, and ironic.
I am real!
Maybe you're the one who's not real.
Maybe you're just an artifact produced by a particular way of looking at the data.
Did you ever consider that?
Somehow I feel like this is a little private joke.
Dunning-Kruger finds that people are still learning and want to feel good about it.